Biochemical Journal

Review article

Chimaerins: GAPs that bridge diacylglycerol signalling and the small G-protein Rac

Chengfeng Yang, Marcelo G. Kazanietz

This article has a correction. Please see:

Abstract

Chimaerins are the only known RhoGAPs (Rho GTPase-activating proteins) that bind phorbol ester tumour promoters and the lipid second messenger DAG (diacylglycerol), and show specific GAP activity towards the small GTPase Rac. This review summarizes our knowledge of the structure, biochemical and biological properties of chimaerins. Recent findings have established that chimaerins are regulated by tyrosine kinase and GPCRs (G-protein-coupled receptors) via PLC (phospholipase C) activation and DAG generation to promote Rac inactivation. The finding that chimaerins, along with some other proteins, are receptors for DAG changed the prevalent view that PKC (protein kinase C) isoenzymes are the only cellular molecules regulated by DAG. In addition, vigorous recent studies have begun to decipher the critical roles of chimaerins in the central nervous system, development and tumour progression.

  • chimaerin
  • diacylglycerol (DAG)
  • phorbol ester
  • protein kinase C (PKC)
  • Rac GTPase-activating protein (RacGAP)

INTRODUCTION

The Ras superfamily of small GTPases or small G-proteins is composed of five major families that share similar sequences and biochemical properties: Ras, Rho, Rab, Ran and Arf [1,2]. More than 150 members in the Ras superfamily have been identified or predicted, and studies have shown that they play important roles in diverse biological processes, including development and progression of human cancers. Unlike heterotrimeric G-proteins, small G-proteins are monomeric. Rho GTPases have been widely studied because they are critical regulators of cellular signalling mediated by GPCRs (G-protein-coupled receptors) and tyrosine kinase receptors. Among more than 20 members of the Rho GTPases, Rac1, Cdc42 (cell division cycle 42) and RhoA have been the most extensively characterized.

Rac plays essential roles in the control of actin cytoskeleton organization, migration, metastasis, transformation, gene expression and cell-cycle progression [36]. Moreover, it has been found that Rac and other Rho GTPases are overexpressed in human neoplasias, such as in breast cancer [3,7]. Three Rac isoforms have been identified: Rac1, Rac2 and Rac3. While Rac1 is ubiquitously expressed, Rac2 is exclusively expressed in haemopoietic cells, and Rac3 is mainly expressed in brain. A Rac1 splice variant designated as Rac1b is mainly expressed in colon and breast cancer tissues [8,9].

Like most other small GTPases, Rac acts as a molecular switch that cycles between an active GTP-bound state and an inactive GDP-bound state. The GTP-bound form of Rac interacts with a number of downstream effectors, including Pak, IRSp53 (insulin receptor substrate p53), Nap125 and PIR121 (p53-inducible mRNA 121) [4,1012]. The activity of Rac isoforms and other GTPases is mainly regulated by three classes of proteins: (i) GEFs (guanine-nucleotide-exchange factors), which promote the exchange of GDP for GTP and thus activate Rac [13]; (ii) GDIs (guanine-nucleotide-dissociation inhibitors), which mainly bind to GDP-bound Rac and limit Rac's access to GEFs and other regulators, effectively maintaining Rac in the inactive (GDP-bound) state [14,15]; and (iii) GAPs (GTPase-activating proteins), which accelerate the intrinsic GTPase activity of Rac, thus leading to Rac inactivation [16].

Cellular signalling from GPCRs and tyrosine kinase receptors to Rho GTPases is mainly mediated by RhoGEFs [17]. RhoGEFs can be activated either directly by tyrosine kinase receptors or indirectly via tyrosine kinase effectors such as PI3K (phosphoinositide 3-kinase). Studies have found that signals from diverse tyrosine kinase receptors can converge on the same Rho-GEF, and that a single tyrosine kinase receptor can activate multiple RhoGEFs. RhoGEFs can be promiscuous in terms of small G-protein activation (for example, Vav2 can activate Rho, Cdc42 and Rac) or can possess a high degree of selectivity, such as Tiam1 for Rac. RhoGEFs are overexpressed in various human cancers and can act as oncogenes. Among the RacGEF family members, Tiam1 and Vav isoforms have been the best characterized. Tiam1 is highly expressed in invasive breast cancer cells and tissues, contributing to cancer progression and metastasis [18]. Vav1 plays critical roles in pancreatic tumorigenesis and is required for the optimal proliferation and survival of pancreatic cancer cells [19]. These findings have made it clear that an imbalanced regulation of Rac activity by RacGEFs and/or RacGAPs may contribute to the development and progression of human cancers.

RhoGAPs

The first GAP for a small G-protein, p120RasGAP, was identified in 1988 [20,21]. Since then, a large number of GAPs for small GTPases have been identified. By searching the human genome, Bernards [22] has predicted that there are 173 human genes encoding GAP-related proteins. Although there are at least 70 RhoGAPs among this large group, only 12 classes have been shown to have GAP activity towards Rac: 3BP-1 [SH3 (Src homology domain 3)-binding protein 1], ABR [active BCR (breakpoint cluster region)-related protein], ARHGAP9 (RhoGAP9, MGC12959), BCR, chimaerins, MgcRacGAP (male germ cell RacGAP) (RacGAP-1, KIAA1478), oligophrenin-1, p190-B (ARHGAP5, RhoGAP5), p50 RhoGAP (Cdc42RhoGAP), PARG1 (protein tyrosine phosphatase-like 1-associated RhoGAP1), RALBP1 (RalA-binding protein 1) (RLIP76) and RICH-1 (RhoGAP interacting with CIP4 homologues 1) (NADRIN) (reviewed in [22,23]). These proteins not only contain a RhoGAP domain, but also have additional structural elements that confer unique regulatory properties, such as specific protein–protein or protein–lipid interactions. A comparison of domain structures of RhoGAPs is depicted in Figure 1. The GAP activity of these proteins plays a critical role in many biological processes. One example of the importance of a GAP protein is that of oligophrenin-1. Studies have shown that oligophrenin-1, which is absent in a family affected with X-linked mental retardation, is required for dendritic spine morphogenesis [24,25]. While expression of a full-length oligophrenin-1 in PC12 (pheochromocytoma) cells decreases the levels of active RhoA, Cdc42 and Rac, only expression of a constitutively active RhoA mutant (RhoAV14) reduces spine length and density that mimics the effects of oligophrenin-1 knockdown using RNAi (RNA interference) and antisense [25]. Moreover, the Rho kinase inhibitor Y-27632 significantly rescued the effect of oligophrenin-1 knockdown on spine length in hippocampal slices, indicating that oligophrenin-1 affects spine morphogenesis through the inactivation of the RhoA/Rho-kinase pathway.

Figure 1 A brief comparison of domain structures among various RhoGAP proteins

ABR, active BCR-related protein; RALBP1, ralA-binding protein 1; RICH-1, RhoGAP interacting with CIP4 homologues 1. Domain name abbreviations: C1, C1 domain; C2, C2 domain; EH, Eps15 homology domain; P, proline-rich domain; PH, pleckstrin homology domain; Sec14, sec-14 like domain; SH2, Src homology 2 domain; SH3, Src homology 3 domain; ZPH, ZK667.1a/PARG homology domain.

Chimaerins are unique among the RhoGAPs in two aspects: (i) biochemically, chimaerins are the only known RhoGAPs with GAP activity specifically for Rac [2629]; and (ii) structurally, chimaerins also contain a C1 domain similar to those present in PKC (protein kinase C) isoenzymes, a family of serine/threonine kinases that have been widely characterized as the main cellular receptors for the lipid second messenger DAG (diacylglycerol) and the phorbol ester tumour promoters. Studies from our laboratory and others have demonstrated that chimaerins indeed represent a unique class of high-affinity receptors for phorbol esters and DAG (reviewed in [3033]), implying a potential role for chimaerins as DAG effectors upon activation of plasma membrane receptors. In the last few years chimaerins have been widely investigated as RacGAPs at the structural, pharmacological and biochemical levels. Their biological functions have also been the subject of intense studies in several laboratories. In the remainder of this review, we will focus on the chimaerin RacGAPs. We will summarize our current knowledge of their structure, biochemical and functional properties. For more information about other RhoGAPs, readers are referred to other excellent reviews [22,23,34].

CHIMAERIN GENES AND STRUCTURES

Chimaerin genes

The first chimaerin isoform was cloned in 1990 from human brain cDNAs by Louis Lim and his colleagues, and it was named n-chimaerin or neuronal chimaerin [35,36], and later renamed α1-chimaerin. There are at least four members in the chimaerin family: α1-, α2-, β1- and β2-chimaerin. These isoforms are alternatively spliced products from the α- and β-chimaerin genes. The α-chimaerin gene (CHN1) was mapped to chromosome locus 2q31-32.1, and the β-chimaerin gene (CHN2) was mapped to chromosome locus 7p15.3 [37]. Studies have shown that α1-chimaerin is mainly expressed in brain [35]; rat α1-chimaerin mRNA is restricted to neurons, with the highest levels in hippocampal pyramidal cells, granule cells of the dentate gyrus and cortical neurons; and in cerebellum the α1-chimaerin mRNA can only be detected in Purkinje neurons [38]. β1-Chimaerin is mainly expressed in testis [39], α2-chimaerin is highly expressed in brain, particularly in neuronal perikarya, dendrites and axons [40], β2-chimaerin is highly expressed in cerebellum and is mainly expressed in granule cells [27]. However, α2- and β2-chimaerin mRNA can be detected in a wide range of cultured cell lines. β2-Chimaerin has been also found in T-lymphocytes [41]. In lower organisms, α-chimaerin expression was detected zygotically and maternally in zebrafish [42]. A chimaerin-like RacGAP in Drosophila, the product of rotund (rn) locus, has been identified [43]. The 1.7 kb RacGAP mRNA transcript was detected in the imaginal discs and in specific germ line cells of the testes, and it presents 60% homology with mammalian MgcRacGAP [43].

Although the chimaerin genes were cloned more than a decade ago, the mechanisms that control chimaerin expression are not well understood. Dong and Lim [44] reported that the expression of α1-chimaerin in neurons is spatially and temporally regulated at a transcriptional level by neuronal/synaptic activity. In human neuroblastoma SK-N-SH cells, α1- and α2-chimaerin mRNA levels are up-regulated during neuronal-type differentiation. KCl-induced membrane depolarization also up-regulates α1-chimaerin mRNA expression in SK-N-SH cells. In addition, the expression of α1-chimaerin was also found to increase in response to hypotonic osmolarity changes [44]. Recent studies confirmed that α1-chimaerin protein levels increase in response to synaptic activity [45]. Rat α1-chimaerin mRNA became detectable in embryonic brain at day 15 and increased in amount postnatally from birth to 20 days after birth, coincident with cellular differentiation and synaptogenesis [38]. A recent study showed that levels of rat α1-chimaerin protein in hippocampal tissue gradually increased during the first 2 weeks of postnatal development, but α2-chimaerin protein levels peak at postnatal day 6 and declined thereafter [45]. Interestingly, an early study found that, in male canaries examined during the song season, α1-chimaerin (named HAT-2 in that publication, 96% protein sequence identical with human α1-chimaerin) mRNA is enriched in forebrain and differentially expressed in the song control circuit [46]. Within the canary forebrain, α1-chimaerin mRNA is notably lower in fibre tracts and in song control nuclei HVC (higher vocal centre), RA (robust nucleus of the archistriatum) and lateral MAN (magnocellular nucleus of the anterior neostriatum), but more abundant in a fourth song nucleus, Area X, compared with the regions in which nuclei sit [46]. However, how the expression of other chimaerin isoforms is regulated is currently not known. Accumulating evidence indicates that chimaerins are down-regulated in some types of cancers [47,48], pointing to the need for deciphering the mechanisms that control their gene expression. Interestingly, studies on the α1-chimaerin promoter region revealed that no TATA box, Sp1 (specificity protein 1)-binding site or initiator motif has been found; instead, it was observed that a CCAAT box located in the proximal promoter region is essential for promoter activity [49]. A 70-bp negative regulatory element in the 5′-untranslated region of exon 1 of α1-chimaerin was found, and deletion of this regulatory element increased the promoter activity 5–6-fold [50]. No studies have yet been carried out to characterize the promoters of other chimaerin isoforms.

Structure of chimaerin isoforms

Chimaerins were originally defined as a ‘chimaera’ between the C1 domain of PKC isoenzymes and the GAP domain of the BCR, a protein involved in the translocation of Philadelphia chromosome in chronic myelogenous leukaemia. All four chimaerins have a C-terminal GAP domain and a single PKC-like C1 domain. α2- and β2-chimaerins also have an additional N-terminal SH2 domain. The domain structures in β2-chimaerin and corresponding biochemical properties are shown in Figure 2.

Figure 2 The domain structures of β2-chimaerin and their corresponding biochemical properties

The C1 domain in β2-chimaerin binds phorbol esters and DAG with high affinity. A representative Scatchard plot for [3H]PDBu binding is shown. The GAP domain in β2-chimaerin has specificity for the small G-protein Rac. A representative in vitro RacGAP shows that increasing concentrations of β2-chimaerin accelerate GTP hydrolysis from Rac.

The chimaerin C1 domain

C1 domains are small structural units of ∼50 amino acids originally defined as lipid-binding modules in PKC isoenzymes. They are also present in several other signalling molecules such as PKDs (protein kinase Ds), RasGRPs, DGKs (diacylglycerol kinases), c-Raf and Vav [51]. C1 domains can be classified into ‘typical’ and ‘atypical’ [51,52]. Typical C1 domains bind phorbol esters and DAG. They have unique patterns of ligand recognition owing to subtle structural differences. The C1 domain in chimaerins is ∼40% homologous with C1 domains in PKC isoenzymes and possesses all the structural requirements for phorbol ester binding, including the motif HX12CX2CX13/14-CX2CX4HX2CX7C, characteristic of other typical C1 domains. Modelling studies determined that the C1 domain has a globular structure with a groove on top, which is the binding site for phorbol esters. Characteristic hydrophobic amino acids that participate in membrane insertion are present in the chimaerin C1 domain [28]. As chimaerin gene splicing occurs upstream of the C1 domain, spliced products from each gene have identical C1 domains [27,37,39]. The C1 domains in α- and β-chimaerins are almost 94% identical.

The C1 domain in chimaerins is crucial for the positional regulation of the protein. Chimaerins redistribute to membranes in response to phorbol ester activation (see below). For β2-chimaerin, a series of site-directed mutagenesis and deletional analyses confirmed that the C1 domain is essential for its ability to bind phorbol esters and to translocate to membranes [28,53]. Mutation of the essential Cys246 to alanine in β2-chimaerin, which disrupts the overall structure of the C1 domain, impairs the ability of the protein to relocalize in response to PMA or other phorbol ester analogues [28]. This analogy with PKC isoenzymes led to the hypothesis that chimaerins could be effectors for DAG generated in response to receptor activation.

The chimaerin GAP domain

The GAP domain in chimaerins is composed of ∼150 amino acid residues and it is highly homologous with the BCR GAP domain [35,54]. The GAP domains of α- and β-chimaerins share ∼77% identity [39]. β2-Chimaerin has been shown to preferentially accelerate the hydrolysis of GTP from Rac with a ∼50-fold higher rate than from Cdc42. No activity for Rho could be detected [28,29]. The RacGAP activity of chimaerins is sensitive to the lipid environment, specifically to the presence of phorbol ester/DAG, PS (phosphatidylserine) and PA (phosphatidic acid) [26].

The chimaerin SH2 domain

α2- and β2-chimaerins have N-terminal SH2 domains that share ∼82% homology [27]. These SH2 domains contain a glutamic acid residue at the start of the domain instead of the tryptophan residue commonly found in other SH2 domains. However, the conserved arginine residues and the phosphotyrosine-binding sequence are preserved. Compared with other SH2 domains, such as the one in Src, α2- and β2-chimaerin SH2 domains have a shorter helix 2 [27]. Currently, little is known about the biological roles of these domains in chimaerins. While studies have shown that the SH2 domain of RasGAP binds to various growth factor receptors [55], it is not known whether the SH2 domains of α2- and β2-chimaerins have similar properties. Interestingly, previous studies have demonstrated that the α2-chimaerin SH2 domain interacts with other proteins, and that this interaction is required for neuritogenesis [40,56]. Expression of α2-chimaerin with a mutation in the SH2 domain (N94H) in NGF (nerve growth factor)-stimulated PC12 cells produced an α1-chimaerin-like phenotype and inhibited neurite formation, suggesting that the chimaerin SH2 domain must associate with proteins that are critical for neuronal differentiation [40].

The three-dimensional structure of β2-chimaerin

A striking progression in our understanding of the regulation and function of chimaerins has been achieved with the elucidation of the three-dimensional structure of β2-chimaerin [57] (Figure 3). This is so far the only structure of a full-length protein with a phorbol ester-responsive C1 domain that has been solved (no full-length PKC structures are yet available). Data from crystallographic analysis revealed that the C1 domain forms extensive hydrophobic contacts with the GAP domain and the SH2 domain. The N-terminus of β2-chimaerin protrudes into the active site of the RacGAP domain, sterically blocking Rac binding. The DAG-binding site in the C1 domain is buried by contacts with the N-terminus, the SH2 domain, the RacGAP domain and the SH2-C1 linker region. These structural characteristics imply that the C1 domain in β2-chimaerin is inaccessible to DAG or phorbol esters when the protein is in this conformation, arguing for the need of a conformational change that exposes the C1 domain in order to bind DAG. This may explain why a higher concentration of phorbol ester is required to induce β2-chimaerin translocation to membranes in cellular models when compared with PKCs [28,58]. The prediction is that lipid binding to the C1 domain requires the co-operative dissociation of these intramolecular interactions, allowing the N-terminus to move out of the active site and enzyme activation. These structural observations were confirmed by mutagenesis studies, in which a series of β2-chimaerin mutants that destabilize the inactive conformation were generated. Remarkably, when these mutants were tested for their ability to translocate from the cytosolic fraction to membranes in response to PMA, it was found that they respond much more readily to PMA than wild-type β2-chimaerin. The mutant I130A-β2-chimaerin is particularly sensitive to PMA-induced translocation and has ∼100-fold lower EC50 for the phorbol ester than the wild-type β2-chimaerin. Interestingly, the EC50 for PMA-induced translocation of I130A-β2-chimaerin is nearly equal to that for PKCα. The RacGAP activity of I130A-β2-chimaerin is also enhanced. Whereas wild-type β2-chimaerin reduces Rac-GTP levels by ∼25% in COS-1 cells, the mutant I130A reduces Rac-GTP levels by ∼80% [57].

Figure 3 Three-dimensional structure of β2-chimaerin

(A) A model for the association of β2-chimaerin to membranes and Rac. Domains are coloured red (SH2), blue (C1), green (RacGAP) and grey (linkers), and Rac is shown in yellow. (B) Molecular surface of inactive β2-chimaerin. Hydrophobic surface is coloured green, basic surface is blue, acidic surface is red and uncharged polar surface is white. (C) Molecular surface of the model of the active membrane-bound β2-chimaerin, coloured as in (B) and viewed looking directly out of the membrane. Reprinted from Cell 119, Canagarajah, B., Coluccio Leskow, F., Ho, J. Y., Mischak, H., Saidi, L. F., Kazanietz, M. G. and Hurley, J. H., ‘Structural mechanism for lipid activation of the Rac-specific GAP, β2-chimaerin’, pp. 407–418, Copyright 2004, with permission from Elsevier.

Based on these findings from the three-dimensional structure, a model for β2-chimaerin activation has been proposed [51]. The inactive protein resides in the cytoplasm; when cells are stimulated with phorbol esters or growth factors trigger generation of DAG, a conformational change occurs and the C1 domain is exposed, an event that facilitates ligand binding, membrane insertion, association with Rac and activation of chimaerin RacGAP activity. Although this model explains how the enzyme activity of chimaerin is activated, it does not tell us how the conformational changes occur. The prediction is that other events, such as post-translational modifications (phosphorylation) or interactions with yet unidentified proteins, must be required to facilitate the transition to the open conformation. At this stage, it is unclear what role the SH2 domain plays in chimaerin activation. It is conceivable that SH2 domain interaction with tyrosine phosphorylated proteins, including receptors, contributes to the process of chimaerin activation and probably co-operates with the C1 domain for membrane association. As in subcellular fractionation studies, chimaerin isoforms could be found in membrane and Triton X-100-insoluble fractions [28], it is also conceivable that alternative regulatory mechanisms might exist in discrete intracellular compartments.

CHIMAERINS AS RECEPTORS FOR PHORBOL ESTERS

As chimaerins contain a PKC-like C1 domain, the binding properties of phorbol esters and DAG analogues to chimaerins became the subject of intensive studies. An early report showed that a TrpE–n-chimaerin fusion protein, expressed in Escherichia coli binds the phorbol ester ligand [3H]PDBu (phorbol 12,13-dibutyrate) in vitro with an affinity of ∼30 nM in a phospholipid-dependent manner [36]. This represented a major finding in the field, as it was the first demonstration of phorbol ester binding by a protein unrelated to the PKC family. However, the Kd for the radioligand observed in this study was much higher than those previously reported for PKC isoenzymes or isolated PKC C1 domains, which bind [3H]PDBu in the high-picomolar/low-nanomolar range [59,60]. Subsequent studies by Areces et al. [61] determined that such discrepancy was due to different methodological approaches used for the binding assays, and demonstrated that recombinant α1-chimaerin and PKCα expressed in Sf9 insect cells have indeed indistinguishable binding affinities for [3H]PDBu. As determined for PKCs, phorbol ester binding to α1-chimaerin is absolutely dependent on the presence of phospholipids. The concentrations of PS required for half-maximal [3H]PDBu binding for α1-chimaerin and PKCα in the absence of calcium are almost identical (14 and 12 μg/ml respectively) [61].

A thorough characterization of β2-chimaerin as a phorbol ester receptor showed important binding similarities with PKC isoenzymes, but also striking differences. Scatchard plot analysis revealed that β2-chimaerin binds [3H]PDBu with low-nanomolar affinity in the presence of PS vesicles (Kd ∼1 nM) [58] (Figure 2). Contrasting results were also observed in structure–activity analysis. The ligand thymeleatoxin, an analogue of the second-stage tumour promoter mezerein, showed a marked preference for binding to PKCα compared with β2-chimaerin (∼60-fold) [58]. On the other hand, DAG analogues displayed a slight preference for β2-chimaerin relative to PKCα [53]. Analysis of co-factor-dependence showed that the acidic phospholipid PS is the most effective phospholipid for supporting [3H]PDBu binding. Unlike PKCα, PS-dependence and ligand-binding affinity were not affected by calcium; in this regard, β2-chimaerin resembles the nPKCs (novel PKCs) [58].

Studies in cellular models determined that phorbol esters and DAG analogues induce the translocation of β2-chimaerin from cytosolic (soluble) to membrane (particulate) fractions, as demonstrated by subcellular fractionation and later confirmed by microscopy using GFP (green fluorescent protein)-fused β2-chimaerin [28,53,58,62]. Moreover, phorbol esters also promote the intracellular translocation of α1-, β1- and α2-chimaerins [28,56]. Deletional and mutagenesis studies showed that the redistribution of β2-chimaerin by phorbol esters is entirely dependent on the C1 domain. Structure–activity relationship analysis of translocation using a series of PKC ligands revealed striking differences between β2-chimaerin and PKCα: the mezerein analogue thymeleatoxin does not translocate β2-chimaerin, as expected from its low affinity for this protein, but it translocates PKCα very efficiently [28]. Together, these findings not only suggest that phorbol esters can positionally regulate chimaerins, but also argue for a differential pattern of binding recognition by individual C1 domains. The identification of selective ligands for each phorbol ester receptor class represents an important step in the design of pharmacological tools to dissect their cellular functions.

BIOCHEMISTRY OF CHIMAERINS

As mentioned above, chimaerins have a GAP domain that is specific for Rac [29,54]. Early studies showed that the chimaerin RacGAP activity is regulated by phorbol esters and phospholipids [26,29]. Ahmed et al. [26] reported that the RacGAP activity of α1-chimaerin is stimulated markedly by PS and PA, and that phorbol esters can synergize with PS and PA. On the other hand, LPA (lysophosphatidic acid), phosphatidylinositol lipids (PtdIns, PtdInsP, PtdInsP2) and arachidonic acid inhibit α1-chimaerin RacGAP activity. IC50 values obtained from competition experiments suggest that α1-chimaerin and the mutants without GAP activity bind Rac1 with similar apparent binding constants [54]. Further studies indicated that PMA promotes the association of β2-chimaerin with Rac1 in COS-1 cells [28], and that this association lasts for a longer time if β2-chimaerin is devoid of its RacGAP activity [63].

The specificity of β2-chimaerin GAP activity has been characterized extensively in in vitro and in cellular models [29]. Experiments in cell-free systems have shown that recombinant β2-chimaerin increases GTP hydrolysis from Rac1 in a concentration-dependent manner, but not from Cdc42 or RhoA. These findings are supported by the crystal structure of β2-chimaerin, which predicts that unfavourable steric and electrostatic interactions with Cdc42 or RhoA would negate catalysis toward these GTPases. It is believed that Arg311 in β2-chimaerin acts as the ‘arginine finger’ that reaches into the active site of Rac and directly stabilizes the transition state for GTP hydrolysis [57]. Although acidic phospholipids markedly enhance the catalytic activity of β2-chimaerin in cell-free systems, PMA seems to have no effect. Studies in COS-1 cells, however, revealed that the ability of β2-chimaerin to decrease Rac-GTP levels is potentiated by PMA [29]. The RacGAP activity of β2-chimaerin was also observed in other cell lines, such as MCF-7 and T-47D breast cancer cells [48,64].

CHIMAERINS AS DOWNSTREAM EFFECTORS OF TYROSINE KINASE RECEPTORS AND GPCRs

The presence of a phorbol ester-responsive C1 domain in chimaerins strongly suggested that these RacGAPs are regulated by DAG generated in response to receptor stimulation. This was a relevant issue to be addressed, as phorbol ester responses are not necessarily equivalent to those generated by DAG upon receptor stimulation. A new paradigm has therefore emerged in which DAG can stimulate pathways independently of PKC.

Studies in our laboratory have used the EGFR [EGF (epidermal growth factor) receptor] as a model to assess chimaerin regulation. The rationale behind this choice was the ability of the EGFR to activate pathways that lead both to Rac activation and DAG generation. EGF promotes cell migration and proliferation via Rac, but the mechanisms for attenuating motogenic and mitogenic signals have not been studied extensively. One of the effectors activated by EGFR stimulation is PLCγ (phospholipase Cγ), which is recruited to the receptor by binding to the phosphorylated tyrosine residue at position 992, a docking site on the receptor cytoplasmic tail [65,66]. Activation of PLCγ causes the hydrolysis of PtdInsP2 into the second messengers InsP3 and DAG.

In a recent study, we identified β2-chimaerin as an effector of the EGFR [63]. EGF causes the translocation of β2-chimaerin from the cytoplasm to the plasma membrane (Figure 4). FRET (fluorescence resonance energy transfer) between YFP (yellow fluorescent protein)–β2-chimaerin and CFP (cyan fluorescent protein)–Rac1 was used to show that EGF-mediated translocation of β2-chimaerin causes its association with Rac at the plasma membrane (Figure 5). Interestingly, GAP-deficient β2-chimaerin mutants with an intact C1 domain (for example, the mutant ΔEIE-β2-chimaerin, in which essential amino acids 298–300 in the GAP domain have been deleted) showed enhanced translocation and sustained association with Rac1 in the FRET assays. RNAi depletion of β2-chimaerin significantly extended the duration of Rac1 activation by EGF, suggesting that β2-chimaerin serves as a mechanism that self-limits Rac activity in response to EGFR activation. It is clear that EGF-mediated activation of β2-chimaerin is DAG-dependent, as determined by the lack of translocation and FRET in PLCγ-depleted cells and upon disruption of the C1 domain by mutagenesis. These findings provided unambiguous evidence for the divergence in DAG signalling downstream of tyrosine kinase receptors via PKC-independent mechanisms. These results also revealed a previously unappreciated scenario in which the activation of the PLCγ–DAG axis by tyrosine kinase receptors directly activates a RacGAP, which in turn attenuates Rac signalling (Figure 6; and an animated version at http://www.BiochemJ.org/bj/403/0001bj4030001add.htm).

Figure 4 Translocation of β2-chimaerin by EGF

COS-1 cells expressing either GFP–ΔEIE-β2-chimaerin (GAP inactive) or GFP–C246A-β2-chimaerin (C1 domain mutant, deficient in phorbol ester binding) were treated with EGF (100 ng/ml). Translocation is not observed in the presence of the PLC inhibitor U-73122 or in the C1 domain mutant. chim, chimaerin.

Figure 5 Association of Rac1 and β2-chimaerin, as demonstrated by FRET

COS-7 cells were transfected with YFP–β2-chimaerin and CFP–Rac1 and then stimulated with EGF (100 ng/ml). FRET was determined in the cell periphery, as described in [63].

Figure 6 Model for the regulation of Rac signalling by chimaerins

EGFR stimulation leads to Rac activation. The PLCγ branch of EGFR leads to DAG generation, translocation of chimaerins and Rac inactivation. Reprinted by permission from Macmillan Publishers Ltd: EMBO Journal [63], copyright 2006. To see an animated version of this Figure, go to http://www.BiochemJ.org/bj/403/0001bj4030001add.htm.

GPCRs can activate PLCβ and DAG generation via the heterotrimeric Gq protein. Recent studies using CXCL12 (also known as stromal-cell-derived factor 1α or SDF-1α), a growth-stimulating chemokine for B-cell progenitors [67], provided the first evidence that GPCRs can also modulate chimaerin activity. CXCL12 promotes the translocation of GFP–β2-chimaerin in Jurkat cells [41]. The modulation of β2-chimaerin by CXCL12 is DAG-dependent. The ability of GPCRs to regulate chimaerin localization via DAG has also been demonstrated for the mAChR1 (muscarinic acetylcholine receptor 1), a cell-surface receptor that couples to Gq–PLCβ. These studies show that mAChR1 stimulation recruits α1-chimaerin to the plasma membrane of cultured hippocampal neurons. The effect occurs in a matter of seconds and depends entirely on DAG generation and the α1-chimaerin C1 domain [45].

INTERACTIONS OF CHIMAERINS WITH OTHER PROTEINS

Several studies have provided clear evidence for the existence of chimaerin-interacting proteins, suggesting that they may be part of multiprotein complexes. Tmp21-I (p23), a transmembrane protein localized in the cis-Golgi network and involved in intracellular trafficking, was identified as a chimaerin-binding protein in a yeast-two hybrid screening using β2-chimaerin as a bait [62]. The association of β2-chimaerin with Tmp21-I is promoted by phorbol esters in a PKC-independent manner. A deletional analysis determined that chimaerins require an intact C1 domain for their interaction with Tmp21-I. These findings have important mechanistic implications, as they reveal a dual role for the chimaerin C1 domain, both as a module for lipid recognition and for protein–protein interaction [51]. Since chimaerins show a perinuclear localization and co-localize with a Golgi network marker in response to phorbol ester treatment, the identification of a Golgi/endoplasmic reticulum protein as a chimaerin partner was not unexpected [28]. These studies also raised the possibility that chimaerins might be involved in the control of transport mechanisms. Interestingly, a large pool of Rac1 is also localized to the perinuclear region [68].

Other proteins interacting with chimaerins have also been isolated. Qi et al. [69] identified the p35 activator of Cdk5 (cyclin-dependent kinase 5) as an α-chimaerin-interacting protein in a yeast two-hybrid screening. α-Chimaerin, p35 and Cdk5 could be co-immunoprecipitated from HeLa cells, and both the Cdk5 kinase activity and the α-chimaerin GAP activity are retained in the protein complex. Since Cdk5 in association with its neuronal activator p35 regulates neurocytoskeletal dynamics, it has been proposed that the association of α-chimaerin with Cdk5–p35 may be required for their co-ordinated involvement in the remodelling of neuronal actin filaments. Moreover, Brown et al. [56] demonstrated that active α2-chimaerin interacts with CRMP-2 (collapsin-response mediator protein-2) and Cdk5–p35 through its SH2 and GAP domains respectively. Together, these findings suggest that the α-chimaerin RacGAP activity plays an important role in regulating the dynamics of neurite outgrowth. More recently, Van de Ven et al. [70] reported that α1-chimaerin is present in dendrites and spines, and that it binds to the NMDA (N-methyl-D-aspartate) receptor NR2A subunit in a phorbol ester-dependent manner. Overexpression of α1-chimerin in cultured hippocampal neurons inhibits the formation of new spines and removes existing spines. The association of α1-chimaerin with the NMDA receptor is required for its ability to modulate spine number, as a α1-chimaerin mutant that binds weakly to NR2A fails to decrease spine density. The α1-chimerin effect depends on its functional GAP domain. The identification of chimaerin-interacting proteins in neuronal models fits well with the observation that chimaerins are highly expressed in the nervous system and with the proposed roles they may play in regulating functions in the CNS (central nervous system).

BIOLOGICAL FUNCTIONS OF CHIMAERINS

Compared with the advances in understanding the biochemical and pharmacological properties of chimaerins, relatively little is known about their biological functions. Since chimaerins are highly expressed in brain, it is expected that they should play important roles in regulating Rac-mediated responses in the CNS. The well-established role for Rac and its effectors on cell proliferation and motility [6] as well its documented role as a Ras effector [7173] suggest potential roles for chimaerins in human cancer development and progression as well.

Chimaerins and the nervous system

Not long after the cloning of α1-chimaerin and its characterization as a RacGAP, Lim's group reported that α1-chimaerin co-operates with Rac1 and Cdc42Hs (Homo sapiens Cdc42) to induce the formation of lamellipodia and filopodia, resembling natural morphological events occurring at the leading edge of fibroblasts and neuronal growth cones [74]. Interestingly, these effects of α1-chimaerin were inhibited by dominant-negative Rac1 (N17Rac1) and Cdc42Hs (N17Cdc42Hs) mutants respectively. An α1-chimaerin mutant devoid of GAP activity and with no p21-binding capability was ineffective in inducing these morphological changes. However, an α1-chimaerin mutant with impaired GAP activity alone was still effective. These findings indicate that, paradoxically, in addition to down-regulating the activity of Rac, α1-chimaerin is also capable of affecting the morphology of neuronal cells independently of its RacGAP activity. No other studies have proposed that chimaerins could have a role in Rac activation, and the physiological relevance of these findings still needs to be validated.

As mentioned above, α-chimaerins have been implicated in the regulation of the dynamics of neurocytoskeleton organization and growth cone guidance through interaction with CRMP-2 and Cdk5–p35, effects that require an intact GAP domain and SH2 domain [56,69]. Indeed, Van de Ven et al. [70] reported that α1-chimaerin is capable of modulating dendritic spine density by binding to synaptic NMDA receptors and locally inactivating Rac1. Overexpression of α1-chimaerin in Purkinje cells in organotypic slice cultures generated dramatic alterations in dendritic morphology [45]. Morphometric analysis revealed a ∼50% reduction in dendritic length and in the number of branch points, with most severe effects in higher-order branches. Both DAG-binding and RacGAP activities of α1-chimaerin were required for the induction of the pruning of dendritic spines and branches. On the other hand, overexpression of α2-chimaerin did not induce dendritic pruning, but it did significantly increase process outgrowth, suggesting a differential role for the SH2 domain in α2-chimaerin. Experiments using RNAi defined further the involvement of α1-chimaerin in regulating dendritic morphology, as they revealed that α1-chimaerin depletion increases protrusive activity from the dendrite. Although the total number of dendritic protrusions was not changed, the number of normal headed spines per unit length of dendrite was decreased by ∼50% when compared with control cells. These findings indicate that α1-chimaerin plays an important role in limiting the expansion of dendritic protrusions and thereby contributes to the normal development of dendritic arbors [45].

The role of the α2-chimaerin SH2 domain in neuronal function has also been investigated. Transient expression of α2-chimaerin, but not α1-chimaerin, in N1E-115 neuroblastoma cells leads to neurite formation [40]. α2-Chimaerin transfectants generate neurites independent of serum stimulation. In contrast, transfection with α2-chimaerin mutated in the SH2 domain (N94H) inhibits neurite formation in NGF-stimulated PC12 cells, indicating a role for α2-chimaerin in neuritogenesis for which its SH2 domain is indispensable [40].

Chimaerins and development

Chimaerins are highly expressed in the rat embryonic nervous system [27,40] suggesting a potential role during development. Recently, in our laboratory, we have developed a zebrafish model in order to provide new insight into the role of chimaerins in development [42]. The chn1 gene product expresses the three modules present in α2- and β2-chimaerins (SH2, C1 and GAP domains), and it is highly homologous with the mammalian α2-chimaerin isoform (Figure 7A). A biochemical analysis of chn1 revealed that, like the mammalian isoforms, it possesses RacGAP activity and phorbol ester-binding capability (Kd [3H]PDBu ∼2 nM).

Figure 7 The zebrafish homologue of α2-chimaerin (chn1): a regulator of epiboly progression

(A) Phylogenetic analysis of chimaerin genes. (B) Injection of a morpholino that depletes chn1 (MO1) or a mismatched morpholino (miMO) was carried out at a one-cell stage. ntl and pax2.1 expression in embryos injected with either miMO or MO1. Embryos were staged and kept at 28 °C until wild-type siblings reached 90% epiboly. Dorsal (top), lateral (second from top) and vegetal (third from top) views are shown. The bottom panel shows DAPI (4′,6-diamidino-2-phenylindole) staining of embryos injected with either miMO or MO1. Embryos were staged and kept at 28 °C until wild-type siblings reached 90% epiboly. (C) chn1 knockdown leads to gain of Rac function and changes in cell distribution along the anteroposterior axis. This is probably due to a dysbalance in cell movements that results in fewer cells reaching the axis with the consequent accumulation of cells in the tailbud. Reprinted from [42], with permission. © 2006 National Academy of Sciences, U.S.A.

A temporal analysis of expression revealed that the chn1 transcript can be detected in eight-cell-stage zebrafish embryos and that it is widely distributed during the cleavage, blastula and gastrula periods. During the segmentation period, the expression of chn1 is restricted to the neural tissue, and, by the pharyngula period, the protein is highly expressed in brain. At the larval stages, chn1 is also present in liver, gut, pancreas and pharyngeal structures, coinciding with the expression pattern of its putative target, zebrafish Rac1 (rac1). chn1 morpholino knockdown embryos displayed severe abnormalities, including the development of round somites, lack of yolk extension and a kinked posterior notochord. These zebrafish morphants showed Rac hyperactivation and progressed faster through epiboly (Figure 7B), leading to tailbud-stage embryos that had a narrowed axis and an enlarged tailbud with expanded bmp4 and shh expression [42].

Mutational studies demonstrated that the lack of chn1 RacGAP activity in the YSL (yolk syncytial layer) is the main cause of the defects in morphogenetic movements. While injection of the GAP-domain-deficient mutant ΔEIE-chn1 mRNA into the YSL of morpholino knockdown embryos was unable to rescue the phenotype, a significant rescue was observed when mRNA encoding the RacGAP domain of mammalian β2-chimerin (β-GAP) was injected into the YSL. These studies not only reveal a crucial role for chn1 in early development, but also implicate Rac as a key regulator of morphogenetic movements during zebrafish epiboly (Figure 7C). In addition, they suggest a role for DAG in the control of early developmental processes.

Chimaerins and cancer

It is now widely accepted that Rac and other Rho GTPases play critical roles in the regulation of cell morphology and movement, invasion, metastasis, proliferation and malignant transformation, which are all crucial events in cancer development and progression [7577]. Rac1 and its spliced variant Rac1b are highly expressed and hyperactive in human cancers [79,78]. Naturally, down-regulation of Rac activity has been explored as a strategy for cancer therapy [7981]. It is reasonable to hypothesize that chimaerins or other RacGAPs possess tumour-suppressor capabilities, and accumulating evidence supports this hypothesis.

The potential link between chimaerins and cancer was first investigated in astrocytoma [47]. Human astrocytomas are the most common primary CNS tumours, but the molecular mechanisms leading to the malignant transformation from low- to high-grade tumours are not well understood. Yuan et al. [47] reported that β2-chimaerin is differentially expressed in brain tumours, with high expression levels detected in normal brain and low-grade astrocytoma and low expression levels detected in malignant high-grade gliomas. These findings support a role for β2-chimaerin as a tumour-suppressor gene and implicate that dysregulation of Rac activity may play a role in the progression of human brain cancers. Likewise, the transcript levels of β2-chimaerin in human breast cancer cells are significantly lower than in immortalized normal breast cells. β2-Chimaerin mRNA levels in human breast cancer tissues are also significantly lower than those in paired normal breast tissues from the same patients, supporting a role for chimaerins in tumour suppression [48,64,82].

Stable transfectants in mouse mammary carcinoma cells expressing the β2-chimaerin GAP domain have reduced proliferation rates and invasiveness capability in vivo [82]. Ectopic expression of β2-chimaerin induces cell-cycle arrest in G0/G1 and inhibits the proliferation of MCF-7 breast cancer cells in a RacGAP-dependent manner. Reduction in Rac-GTP levels by expression of the β-GAP chimaerin domain in MCF-7 cells correlates with decreased expression of cyclin D1, reduced pRb (retinoblastoma family protein) phosphorylation levels and inhibition of cell-cycle G1/S transition. Moreover, the effects of β2-chimaerin on MCF-7 cell proliferation can be rescued by constitutively active Rac (V12Rac1) but not RhoA (V14RhoA) [48,64].

Recent studies have pointed to a crucial role for Rac as a mediator of growth factor responses in breast cancer cells [48,64]. EGF and HRG (heregulin β1) cause significant elevations in Rac-GTP levels in breast cancer MCF-7 and T-47D cells and promote breast cancer cell migration and proliferation in a Rac-dependent manner. Unlike EGF, the effect of HRG on Rac activity is sustained. β2-chimaerin inhibits HRG-induced breast cancer cell migration and proliferation through the inactivation of Rac [64], pointing to a role for β2-chimaerin in impairing growth factor-mediated responses that depend on Rac. As Rac mediates growth-factor-induced generation of ROS (reactive oxygen species) [83,84], chimaerins may modulate this effect, an issue that has not yet been explored.

Other functions of chimaerins

In addition to the anti-proliferative and anti-migration actions of β2-chimaerin on cancer cells, a recent study has shown similar effects in vascular smooth muscle cells: Maeda et al. [85] reported that ectopic expression of β2-chimaerin inhibits PDGF (platelet-derived growth factor)-induced Rac activation, thus impairing smooth muscle cell migration and proliferation. This study points to a potential role for chimaerins in human atherogenesis.

Siliceo et al. [41] recently found that β2-chimaerin is expressed in T-lymphocytes. Expression of β2-chimaerin inhibits Rac-GTP levels in T-cells, an effect that depends on its RacGAP activity and requires DAG generation. Interestingly, while β2-chimaerin reduces static adhesion, it enhances CXCL12-dependent migration through receptor-dependent DAG production. Cells expressing the GAP-inactive mutant ΔEIE-β2-chimaerin showed higher levels of Rac-GTP and increased adhesion induced by CXCL12 or PMA compared with cells expressing wild-type β2-chimaerin. Moreover, a C1 domain mutant (F215G) inhibited the effect of wild-type β2-chimaerin on PMA-induced integrin-dependent adhesion, and also prevented the inhibitory effect of wild-type β2-chimaerin on integrin-dependent adhesion following CXCL12 stimulation, demonstrating an absolute requirement of an intact C1 domain for inhibition of β2-chimaerin on cell adhesion.

CONCLUDING REMARKS

Chimaerins represent the first example of RacGAPs regulated directly by the lipid second messenger DAG. Extensive pharmacological and biochemical studies have unambiguously established that these RacGAPs are high-affinity receptors for phorbol ester tumour promoters. These findings, along with the discovery of proteins with phorbol ester-binding capabilities that do not belong to the PKC family, have challenged the traditional view that PKC isoenzymes are the only family of intracellular DAG effectors. It is now clear that multiple effectors convey signals triggered by the generation of DAG or stimulation with phorbol esters. Cross-talk between DAG-regulated effectors has been demonstrated for RasGRP1 and PKC [86], and it is likely that this could also be the case for chimaerins. Indeed, β2-chimaerin becomes phosphorylated in response to receptors that couple to PLCγ and DAG generation (E. M. Griner and M. G. Kazanietz, unpublished work). A challenge is to clarify the signalling pathways and biological functions mediated by individual members of the chimaerin family and their functional interactions using pharmacological and genetic tools. The biochemical and biological evidence that chimaerins are specific RacGAPs is compelling, but only recently has their role in limiting Rac activation via receptor stimulation begun to be elucidated. In addition to EGFR, can other PLCγ-coupled receptors regulate chimaerin activity and stimulate the auto-inhibitory loop for Rac inactivation? Is there a contribution of the SH2 domain in α2- and β2-chimaerin to membrane association? Can additional signals lead to chimaerin activation or sensitize these RacGAPs for the membrane anchoring via DAG binding? These important questions will probably unravel the molecular intricacies of chimaerin activation. A better comprehension of the dynamics of subcellular compartmentalization of chimaerins and other RacGAPs in response to stimuli will have a great impact on our understanding of the molecular regulation of Rac and its effectors in cancer and neuronal function.

Acknowledgments

The laboratory of M.G.K. is supported by grants from NCI (National Cancer Institute) of the NIH (National Institutes of Health).

Abbreviations: BCR, breakpoint cluster region; Cdc42, cell division cycle 42; Cdk5, cyclin-dependent kinase 5; CFP, cyan fluorescent protein; CNS, central nervous system; CRMP-2, collapsin-response mediator protein-2; DAG, diacylglycerol; EGF, epidermal growth factor; EGFR, EGF receptor; FRET, fluorescence resonance energy transfer; GAP, GTPase-activating protein; GEF, guanine-nucleotide-exchange factor; GFP, green fluorescent protein; GPCR, G-protein-coupled receptor; mAChR1, muscarinic acetylcholine receptor 1; MgcRacGAP, male germ cell RacGAP; NGF, nerve growth factor; NMDA, N-methyl-D-aspartate; PA, phosphatidic acid; PARG, protein tyrosine phosphatase-like 1-associated RhoGAP; PC12, pheochromocytoma; PDBu, phorbol 12,13-dibutyrate; PKC, protein kinase C; PLC, phospholipase C; PS, phosphatidylserine; RNAi, RNA interference; SH, Src homology domain; YFP, yellow fluorescent protein; YSL, yolk syncytial layer

References

View Abstract